SAR图像溢油信息的提取和识别算法研究

项目名称： SAR图像溢油信息的提取和识别算法研究

项目编号： No.41271434

项目类型： 面上项目

立项/批准年度： 2013

项目学科： 天文学、地球科学

项目作者： 张渊智

作者单位： 南京信息工程大学

项目金额： 75万元

中文摘要： 近年来，海上溢油事件频发，造成大面积海域污染，不仅使海洋生态环境受到损害，造成海洋生物大量死亡，经济蒙受损失，而且危及人类健康。遥感技术，特别是合成孔径雷达（SAR）已成为当前海上溢油监测的主要手段，SAR图像溢油信息提取和自动识别算法就成为快速准确了解溢油发生的位置和面积、溢油量和扩散趋势等信息的关键。 本项目针对SAR图像上的溢油信息提取中的难点，拟研发一种基于小波变换和水平集的新的快速实用的黑斑检测算法，综合提取黑斑的散射、几何、纹理、上下文和极化等特征，最后通过分类器区分出溢油和似然物，从而形成SAR图像的采集、预处理、黑斑检测、特征提取和分类为主线的溢油信息提取技术流程。此外，在准确识别溢油区域的基础上，通过实验室微波散射信号测量，进一步识别溢油种类和估算油膜相对厚度，为海上溢油监测和应急管理提供科学依据和决策支持。

中文关键词： 溢油；合成孔径雷达；极化；图像信息提取；微波散射特性

英文摘要： Oil spill has recently been leading to large scale of environmental polution to the ocean, and thus causing death to lots of marine organism, large amount of economic losses, and great harm to human beings as well. Remote sensing, especially synthetic aperture radar (SAR), has become the main technique for oil spill monitoring. However, it is still a challenging task to automatically detect the oil spill accurately and fastly. This project aims to develop a new image segmentation algorithm based on wavelet transformation and level set method for automatically and fastly extracting the dark spots from SAR images.Scattering, geometric, texture, context and polorization features will be extracted from the detected dark spots. Then, classification will be conducted to distinguish the dark spots into oil slick and look-alikes. The proposed approach consists of a comprehensive framework including SAR data acquisition, processing, segmentation, classification and validation for the oil leakage extraction. Moreover, comparisons with existing methods are also conducted to test the effectiveness of the proposed algorithm. Additionally, on the basis of oil dection, laboratory experiment over the microwave scattering characteristics of various oil will also be designed to further distinguish different types of oil as well

英文关键词： Oil Spill；SAR；Polarimetry；Image Information Extraction；Microwave Scattering Characteristics